The Download: Google DeepMind AI agent and Montana’s experimental treatments

The news: A bill that allows clinics to sell unproven treatments has been passed in Montana.

There will be the first US hub of experimental medical treatments

According to the news, a bill allowing clinics to sell unproven treatment has been passed in Montana. The legislation allows doctors to apply for a license in order to open an experimental treatment center and recommend and sell therapies that are not approved by FDA to their patients.

— Jessica Hamzelou

Google DeepMind’s new AI agent solves real-world issues better than humans

Google DeepMind used large language models once again to find new solutions to longstanding problems in mathematics and computer science. This time, the firm has demonstrated that its approach is not only able to solve unsolved mathematical puzzles but also improve a variety of important real-world process.

AlphaEvolve is a new tool that uses large language models to generate code for various tasks. LLMs have a reputation for being hit or miss when it comes to coding. AlphaEvolve scores all of Gemini’s suggestions. It discards the bad and tweaks the good in an iterative procedure, until it produces the best algorithm possible. Read the full story. In many cases, results are more accurate or efficient than the best (human-written solutions).

– Will Douglas Heaven

Research cutbacks threaten crucial climate data.

– Casey Crownhart

In the last few weeks there has been a flurry of news about proposed cuts to US science budgets. Researchers and civil servants have warned that budget cuts could lead to the loss of important data that can help us understand our world, and how climate changes are affecting it.

According to my colleague James Temple, long-running US government programmes that monitor the snowpack in the West are being threatened by cuts throughout the US federal government. Carbon dioxide measurements in Hawaii, hurricane prediction tools, and a data base that tracks the economic impacts of natural disasters are also at risk.

All of this has me wondering: What are we losing when data is at risk? Read the complete story. This article is taken from The Spark, MIT Technology Review’s weekly climate newsletter. Sign up to receive it every Wednesday by clickingorhere

The must-reads

I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.

Donald Trump doesn’t like Apple building iPhones on Indian soil
According to the US President, Apple will increase their US production in response. (Bloomberg $)
+Also, he said that India would “literally” charge us no tariffs. (WSJ $)

2 Elon Musk’s Grok chatbot rants about white genocide.
In answer to totally unrelated questions. (FT $)
+ This is not the first time Grok’s responses have been questionable. Grok said it was told to accept white genocide. The Guardian

RFK Jr does not think we should follow his medical advice
This begs the questions: Why is he US Health and Human Services Secretary? (NY Mag$)
+ Kennedy stated that his opinions on vaccinations are irrelevant. He defended the decision to reduce the size of the health department despite protests. The Guardian (19459038)

GM’s new EV can power a truck up to 400 miles
Their lithium manganese-rich cell uses cheaper minerals than those of lithium-ion. (Fast Company – $)
+ Tariffs are bad for batteries. MIT Technology Review

Five Anthropic was accused of using AI-generated proof in a court case
An attorney for Universal Music Group claimed that an expert cited a non-existent source. (Reuters ),
and A court in another case also caught fake AI citations. Ars Technica
+ The AI companies have finally been forced to pay for training data. (MIT Technology Review) (19659035]6 AI will not put radiologists out of a jobs any time soon
Although the technology is helpful, it is unable do everything that trained human experts are capable of. (NYT $)
+ The difficulty of using AI to diagnose cancer. MIT Technology Review

7 The US Defense Department is looking for faster aircraft and missiles.
Startups are more than happy to respond. (WP$)
+ The second phase of military AI is here. MIT Technology Review

A major hurdle has been cleared ahead of the planned launch later this week. (Wired$)

– YouTube will begin inserting ads during crucial moments of videos
Wow! That doesn’t seem at all annoying. TechCrunch (19459038]

]10 Apple’s Vision Pro is a pain
and early adopters regret shelling out $3500 each. (WSJ $)
+ Perhaps the ability to scroll with their eyes will change people’s minds. ($Bloomberg )

Today’s Quote

To say a prof is a monster for using AI in order to create slides “is to me, absurd.”

– Paul Shovlin, an Ohio University professor, reacts to the backlash from students against professors who use AI to create teaching material, according to the New York Times (19459038).

Another thing